• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Question about direct coupling a anode follower into a cathode follower.

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.

G

Member
Joined 2002
As I was shown and taught by Sy (haven't heard from him in a while) a simple cathode follower with a single sided power supply has some limitations because of the grid being referenced to ground and sitting essentially at zero volts. In a direct coupled two stage circuit where an anode follower is direct coupled to a cathode follower does the cathode follower still suffer from those limitations? It would seem to me that it does not as the grid is lifted quite a ways above ground by the plate voltage of the anode follower. Am I looking at this correctly?
 
diyAudio Senior Member
Joined 2002
Hi,

So I could direct couple a 5687 CF to the anode follower of my 5687 preamp and lower the output z significantly.

You mean the other way around, right ?
Does that lower the Zout of the CF ? I don't recall it doing that, can't see how it would really.

How do you think that would sound sonically Frank?

This is the kind of circuit we've been trying to avoid for the past ten years, maybe for the wrong reasons...

Anyway, from previous posts I gather you plan on building a preamp with good sonics and low Zout.
Why not try something with either the 12B4A or EL84/6BQ5 trioded with either choke or resistive load?
While certainly not the last word in Zout it's undoubtedly hard to beat on sound.

If you want lower Zout with that type of tube and still need some gain then you can also try a WOT with this.

Or, go for one like this one if you listen to headphones as well:

Cheers,;)
 

Attachments

  • 12b4a_pre_headph.gif
    12b4a_pre_headph.gif
    9.3 KB · Views: 740
a simple cathode follower with a single sided power supply has some limitations because of the grid being referenced to ground and sitting essentially at zero volts.
Yes, a CF connected this way is not very linear and can not handle large voltage swing.

In a direct coupled two stage circuit where an anode follower is direct coupled to a cathode follower does the cathode follower still suffer from those limitations?
No it doesn't if you allow enough voltage over the cathode resistor, (this has to be larger than in an ordinary CF of course). the output impedance of the CF will be the same but will be shunted by a higher resistance.

Another way of increasing grid voltage and allowable voltage swing is to use 2 resistors in series between cathode and ground and then connect the grid leak to the connection point between those resistors. With this connection you can use capacitor coupling.

In any case if you allow 100V or more between cathode and ground of the CF using one of the above methods the linearity will in practise be the same as when using a CCS in the cathode, the only thing you gain by using a CCS is that you can use somewhat lower supply voltage but you dont really gain anything in linearity.

BTW, please don't call it anode follower, a follower is an amplifier with gain of ~1, call it cathode grounded amplifier that is the usual name.

Regards Hans
 
Which is the same grid-leak-DC-operationage-ism as direct coupling to a plate.

Except for the lower input impedance, if you want to maintain the very high input impedance of the CF the solution with 2 series connected cathode resistors works very well, stability for any alternative is not of any serious concern except if you want to DC couple the output.

Regards Hans
 
You can bootstrap the divider too

I think it is what I have already described... the input impedance of a CF with dual cathode resistors is RG/(1-A*RL/*RL+Rk()where RL is the resistor closest to ground and Rk is the resistor closest to cathode, quite a difference to your variant using a potential divider between B+ and ground... I try never to waste gain or power in resistors....

Regards Hans
 
Not like a few hundred K is much lost impedance. No worse than a light grid leak on any other grounded cathode stage.

I'd set up a voltage divider (from +300V to bias at 150Vk, two 1M resistors would do it), cap couple to cathode and run a resistor to grid. Impedance should be something like Rg/(1-gain) or however you'd express it. Which is electrically identical to the cathode biased method, except the voltage divider loads the output slightly. (It'll notice a gnat flying around the envelope before it notices 500k load, so, yeah.)

I don't know how a triode would tend to behave when self-biased as you advocate, but it doesn't feel like good, stable practice. Certainly wouldn't cut it in direct-coupled circuits like my active preamp design.

Tim
 
I don't know how a triode would tend to behave when self-biased as you advocate, but it doesn't feel like good, stable practice.

Maybe you have to test things in reality to find out what works or not... the circuit i describe I have described is completely stable and the grid cathode voitage is defined by the drop over the cathode resistor, same circuit is for instance used in the Marantz 7, I have used the method described before with good result. I noticed that you have now changed your preferred circuit to include bootstrapping but i fail to see any advantages over the more traditional circuit I described...

Regards Hans
 
Keep stuff easy. First, define what you need the circuit to do in terms of Zout, required swing, and what you expect it to drive. KISS.

In general, I don't like schemes which require big coupling caps, especially big HV coupling caps with lots of dielectric in the way, so I've tended to recommend circuits where the cathode sits at a relatively low voltage. And I really, really like direct coupled outputs- inputs can use nice, small caps because of the high Z. Let's start with defining your requirements and I'll go through a step by step, culminating in a finished design, with a few options about coupling.
 
In general, I don't like schemes which require big coupling caps, especially big HV coupling caps with lots of dielectric in the way, so I've tended to recommend circuits where the cathode sits at a relatively low voltage.

Agreed, but a CF with low voltage between cathode and ground is not very linear, 2 ways to making it more linearis either to use higher voltage say 100V or using a CCS. I have compared these 2 solutions and found no practical benefits with the CCS solution except maybe for less requirements on the power supply.

The variant of CF with dual cathode resistors have the advantage of having very high input impedance, much higher than the grid leak which allow for use of small caps, the output is maybe at 100V or so which is acceptable I think when it comes to finding a suitable cap. If you don't like caps you can use a negative voltage to feed the cathode resistor and put the cathode at 0V making it possible to eliminate the cap or at least relax the requirements on it.

In my own preamp I dont use any coupling caps at all, I use DC coupling, (A SRPP connected to a cathode grounded stage connected to a cathode follower) this works very well with wide bandwidth and low distortion, I use AC and DC feedback to stablise the operating point and gain but I am among those who believe in global feedback if done correctly.

Regards Hans
 

G

Member
Joined 2002
SY said:
Keep stuff easy. First, define what you need the circuit to do in terms of Zout, required swing, and what you expect it to drive. KISS.

In general, I don't like schemes which require big coupling caps, especially big HV coupling caps with lots of dielectric in the way, so I've tended to recommend circuits where the cathode sits at a relatively low voltage. And I really, really like direct coupled outputs- inputs can use nice, small caps because of the high Z. Let's start with defining your requirements and I'll go through a step by step, culminating in a finished design, with a few options about coupling.

Hi Sy. Long time no hear. Is this post in response to what I need to drive my amp or are you anwering someone else? Let's assume that you are responding to my post. I would like to have a preamp with about a gain of 6 - 8 with a output Z of around 100 - 200 ohms. I was reading about the Bottlehead Foreplay and discovered that they direct couple the gain stage to the output stage of the Foreplay. I then started thinking about direct coupling and what it meant in reference to the buffer that you so kindly helped me with a few weeks back. One of the drawbacks of the common CF is that the grid sits at earth potential and it's difficult to get it more than a few volts negative to the cathode without using a negative rail. After thinking about direct coupling a bit I saw that it would sit the grid at a high voltage above earth which would in turn enable you to bias the cathode voltage higher without concern of negative clipping due to the grid being essentially at earth. I may be looking at this wrong. Like that's new but I thought it would enable me to build a relatively simple circuit and still get the performance I am looking for. Am I wrong?
 
No, you're not at all wrong. Floating the grid up to a voltage that's an order of magnitude more than you want to swing will work fine as long as you're willing to deal with getting rid of the inevitable DC on the output. And running my signal through a 10u/200V cap just esthetically bothers me, more so than running it through the 0.01-0.1u cap you'd need to couple into the grid of the output stage if that grid is biased to null out the DC component on the cathode.

But floating up the circuit and using that big output cap is absolutely doable and I can't claim that the methods I prefer (which have the tradeoff of requiring a negative rail) would show any better in a controlled listening test. I'll draw up a couple candidate circuits for you and we can discuss.
 
SY said:
No, you're not at all wrong. Floating the grid up to a voltage that's an order of magnitude more than you want to swing will work fine as long as you're willing to deal with getting rid of the inevitable DC on the output. And running my signal through a 10u/200V cap just esthetically bothers me, more so than running it through the 0.01-0.1u cap you'd need to couple into the grid of the output stage if that grid is biased to null out the DC component on the cathode.

But floating up the circuit and using that big output cap is absolutely doable and I can't claim that the methods I prefer (which have the tradeoff of requiring a negative rail) would show any better in a controlled listening test. I'll draw up a couple candidate circuits for you and we can discuss.

Thank you Sy. I would enjoy that. I look forward to hearing from you.
 
OK, I drew a few things up and now I'm trying to get the #@!% scanner to work! While that's going on, let's take step one and pick a CF tube. That's the easy part- we start with your requirement Zo = 200 ohm. I'm arbitrarily not going to use an output transformer. I like to avoid them where possible for reasons of expense, bulk, and low frequency performance. Galvanic isolation is nice, but if that's what I want, I'd rather do it at the input where the signal level is lower, the transformer load is under my control, and where dodgy grounds from signal sources abound.

OK, then. The source impedance of a simple triode CF (and I'm being arbitrary here, you could use a pentode, too) will look like ~1/gm. So the required gm is 5000 umho or higher.

The required swing for a preamp output will be roughly 4-5 volts peak. And the capacitance of the load will be 500 pF or less- assuming you're not driving hideously long cables. So... if we want to avoid the complication and expense of paralleled tubes and stick to classic tube types, a 6DJ8 family member would be quite adequate. As an alternative, you could also use paralleled 6SN7s at the cost of doubling up tubes. But octals do look cool.

OK, make your choice and we'll go to the next step.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.